Announcing the TAPR "Clock-Block" Clock Synthesizer
Sorry for the advertisement, but I thought this would be of particular
interest to the folks on these lists.
Some time ago I mentioned that I was working on a clock synthesizer
board that could be used (among other things) to get much better
timekeeping for applications like NTP.
The idea is to replace the motherboard crystal with a connection to the
synthesizer, which can be programmed to generate the correct frequency
when driven by a higher quality external oscillator. The end result is
a PC clock that's far more accurate and stable than the original crystal
can provide.
I'm pleased to announce that my project worked out, and TAPR is now
taking orders for the "Clock-Block" as a fully-assembled unit (however,
you'll still have to do some soldering on your motherboard to replace
the crystal with the connection to the Clock-Block).
We are having the first batch manufactured now, and hope to begin
shipment within a couple of weeks. I'm pleased that we'll be able to
offer the assembled unit for a price not much more than I originally
thought a kit version would cost; the circuit uses surface mount parts
that might be challenging for some folks to solder, so offering it
assembled makes life easier for everyone.
There's more information about the Clock-Block, and you can place an
order, at http://www.tapr.org/kits_clock-block.html
John
PS -- if you're not familiar with TAPR (http://www.tapr.org), it's a
non-profit organization that does research and development for the ham
radio and amateur electronics enthusiast community. I've been involved
with the group for many years and have developed several projects for
TAPR, but have no financial interest in the organization or any of its
products other than as a member.
Is your PPS signal too short to trigger your serial port? Is that
what's troubling you, Bunkie?
If so, the new TAPR FatPPS is the answer you've been looking for...
The FatPPS is a DB-9 "dongle" that implements a pulse stretcher. Input
signals as short as 20 nanoseconds produce an output pulse about 30
milliseconds wide (the width can be changed by changing a few component
values). It can be powered from the host computer's serial port, and so
requires no external connections at all. It works with either TTL or
RS-232 level input pulses (the output is always TTL level), and can
invert input or output pulse polarity if needed.
I designed the FatPPS because the ~20 microsecond wide PPS signals from
several of my time sources (like the Z3801A GPSDO, and HP frequency
standards) was too short to reliably work with the serial ports on my
NTP servers. Hopefully, it will be of use to some of you as well.
The FatPPS is available only as a fully assembled and tested unit, and
is shipping now. The price is $44 for TAPR members, and $49 for
non-members.
There are more details, and you can place an order, at
http://www.tapr.org/kits_fatpps.html. You can view the installation and
operations manual (which is still a work in progress) at
http://www.tapr.org/~n8ur/FatPPS_Manual.pdf.
Thanks for the interruption.
John
I've made a such pulse stretcher to render the 1ms
pulse from a CMC Superstar and the 10µS pulse from a
Trimble Lassen SK8 visible via an L.E.D. by using a
simple 555 timer configured in a monostable fashion.
Sure it is not nanosecond precise and i'm not sure AT
ALL it will react to a 20ns pulse, but for a visual
monitoring, it is perfect (pulse stretched to about
30ms).
73 de Normand VE2UM
Montreal, QC Canada
--- John Ackermann N8UR jra@febo.com wrote:
Is your PPS signal too short to trigger your serial
port? Is that
what's troubling you, Bunkie?
If so, the new TAPR FatPPS is the answer you've been
looking for...
The FatPPS is a DB-9 "dongle" that implements a
pulse stretcher. Input
signals as short as 20 nanoseconds produce an output
pulse about 30
milliseconds wide (the width can be changed by
changing a few component
values). It can be powered from the host computer's
serial port, and so
requires no external connections at all. It works
with either TTL or
RS-232 level input pulses (the output is always TTL
level), and can
invert input or output pulse polarity if needed.
I designed the FatPPS because the ~20 microsecond
wide PPS signals from
several of my time sources (like the Z3801A GPSDO,
and HP frequency
standards) was too short to reliably work with the
serial ports on my
NTP servers. Hopefully, it will be of use to some
of you as well.
The FatPPS is available only as a fully assembled
and tested unit, and
is shipping now. The price is $44 for TAPR members,
and $49 for
non-members.
There are more details, and you can place an order,
at
http://www.tapr.org/kits_fatpps.html. You can view
the installation and
operations manual (which is still a work in
progress) at
http://www.tapr.org/~n8ur/FatPPS_Manual.pdf.
Thanks for the interruption.
John
time-nuts mailing list
time-nuts@febo.com
Do You Yahoo!?
Tired of spam? Yahoo! Mail has the best spam protection around
http://mail.yahoo.com
Normand Martel wrote:
I've made a such pulse stretcher to render the 1ms
pulse from a CMC Superstar and the 10µS pulse from a
Trimble Lassen SK8 visible via an L.E.D. by using a
simple 555 timer configured in a monostable fashion.
Sure it is not nanosecond precise and i'm not sure AT
ALL it will react to a 20ns pulse, but for a visual
monitoring, it is perfect (pulse stretched to about
30ms).
Hi Normand --
I don't know of anyone who actually has to deal with a 20ns pulse, but
it was interesting to see what the chip could do. I actually tested
down to 8ns and it worked there, but I like to be a bit conservative
when quoting performance.
The main raison d'etre for the FatPPS is to deal with the typically 20us
pulse width that comes from HP clocks. That's just a bit too short to
reliably trigger a serial port. I also wanted to be able to easily turn
a negative-going RS-232 pulse (which is what comes out of the Z3801A
after the "standard" RS-232 modification) into a positive going pulse to
feed the high-resolution Soekris net4501 time server.
John
If you are using the pulse to estimate the transfer function of a system
and you do not want to have to factor out the squared off pulse, the
shorter the better. This is to be balanced against spreading the energy
out so much that you have to nearly overload the measured system with
the pulse or have so little energy in your region of interest that the
noise floor rises to greet you and ruin you! A pulse train helps but
then there are other ambiguities to deal with as well as the need to
decide on the pulse repetition frequency that does you the most good.
I have stolen and adapted various things to do this from Clark and his
VLBI work.
Bob McGwier
John Ackermann N8UR wrote:
Normand Martel wrote:
I've made a such pulse stretcher to render the 1ms
pulse from a CMC Superstar and the 10µS pulse from a
Trimble Lassen SK8 visible via an L.E.D. by using a
simple 555 timer configured in a monostable fashion.
Sure it is not nanosecond precise and i'm not sure AT
ALL it will react to a 20ns pulse, but for a visual
monitoring, it is perfect (pulse stretched to about
30ms).
Hi Normand --
I don't know of anyone who actually has to deal with a 20ns pulse, but
it was interesting to see what the chip could do. I actually tested
down to 8ns and it worked there, but I like to be a bit conservative
when quoting performance.
The main raison d'etre for the FatPPS is to deal with the typically 20us
pulse width that comes from HP clocks. That's just a bit too short to
reliably trigger a serial port. I also wanted to be able to easily turn
a negative-going RS-232 pulse (which is what comes out of the Z3801A
after the "standard" RS-232 modification) into a positive going pulse to
feed the high-resolution Soekris net4501 time server.
John
--
AMSAT Director and VP Engineering. Member: ARRL, AMSAT-DL,
TAPR, Packrats, NJQRP, QRP ARCI, QCWA, FRC. ARRL SDR WG Chair
"If you board the wrong train, it is no use running along the
corridor in the other direction. " - Dietrich Bonhoeffer
Received two of the subject units, one built in 96 and one in 97.
The dates are the first two numbers in the serial number.
Took the 96 apart and learned that the LPRO-101 is not physically
connected to the main board, except through a pin connector. When
the unit is turned upside down, you can see a pattern of 6 screws
in two rows of three. These secure the LPRO. The rest of the screws
hold the main board. Do not turn the unit upside down but stand it
on its rear fins as you remove the screws. This keeps the LPRO from
dropping away and bending the pins.
The LPRO has a thermal plate that provides good conductivity to the
case, so it can dissipate about 15 watts. There is no thermal goo
between the plate and the box, but the plate may be stuck to the box.
Put a screw in a hole and push on it to release the LPRO. Then, if
you have removed the 10 screws that hold the panel to the box, it
should all lift out easily, with the LPRO hanging by its connector.
I'd recommend laying the box down in its normal position and sliding
the innards out horizontally.
Imagine my surprise when what I'd learned didn't work on the 97 box.
That assembly has a clamp that holds the LPRO in place, so pushing on
a screw doesn't work. There are four screws on the back, between the
fins. Loosen them to loosen the clamp, but don't remove them.
Reassembled both - you don't want to run the LPRO without a heat sink.
The 97 unit drew 1.4 amps at 24 VDC at startup. The NO GPS and Fault
lights were lit. The current dropped to .7 amp but the Fault light was
still lit. Got 10 MHz at the J1 test point but nothing at the 10 MHz
output. Went away for an hour, came back to find the ON light, got 10
MHz at J2, didn't check 15 MHz. The 96 unit had the same behavior, but
the 10 MHz was more accurate. The box got quite warm, but you could
keep your hand on it. (How's that for quantitative?)
I have two Z3801s and two Racal-Dana model 1992 counters set to Phase
A-B.
One counter has been measuring a 4040 cesium box against one of the GPS
receivers. The other counter measures the phase difference between the
Z3801s. With the 96 unit in place of the 4040, phase rotation was 220
degrees in 20 minutes. The GPS comparison doesn't rotate, but varies
between 50 and 70 degrees. I consider that an indication that the
receivers
are OK. The holdover accuracy varies between 0.5 and 5 microseconds,
depending on when the crystal last cracked.
If I did the math right, a 360 rotation will take 2000 seconds. This
means
the LPRO is about 5x10E-11 off from GPS, no?
Two other questions:
What is the BITE pin on the LPRO? Some Built-In Test?
What is the protocol for the data at J6? Can it be the same as the
Z3801?
Bill Hawkins
I have a '97 box and the disassembly is as you
described for the '96 box, except I only have 9 screws
on the bottom.
--- Bill Hawkins bill@iaxs.net wrote:
Received two of the subject units, one built in 96
and one in 97.
The dates are the first two numbers in the serial
number.
Took the 96 apart and learned that the LPRO-101 is
not physically
connected to the main board, except through a pin
connector. When
the unit is turned upside down, you can see a
pattern of 6 screws
in two rows of three. These secure the LPRO. The
rest of the screws
hold the main board. Do not turn the unit upside
down but stand it
on its rear fins as you remove the screws. This
keeps the LPRO from
dropping away and bending the pins.
The LPRO has a thermal plate that provides good
conductivity to the
case, so it can dissipate about 15 watts. There is
no thermal goo
between the plate and the box, but the plate may be
stuck to the box.
Put a screw in a hole and push on it to release the
LPRO. Then, if
you have removed the 10 screws that hold the panel
to the box, it
should all lift out easily, with the LPRO hanging by
its connector.
I'd recommend laying the box down in its normal
position and sliding
the innards out horizontally.
Imagine my surprise when what I'd learned didn't
work on the 97 box.
That assembly has a clamp that holds the LPRO in
place, so pushing on
a screw doesn't work. There are four screws on the
back, between the
fins. Loosen them to loosen the clamp, but don't
remove them.
Reassembled both - you don't want to run the LPRO
without a heat sink.
The 97 unit drew 1.4 amps at 24 VDC at startup. The
NO GPS and Fault
lights were lit. The current dropped to .7 amp but
the Fault light was
still lit. Got 10 MHz at the J1 test point but
nothing at the 10 MHz
output. Went away for an hour, came back to find the
ON light, got 10
MHz at J2, didn't check 15 MHz. The 96 unit had the
same behavior, but
the 10 MHz was more accurate. The box got quite
warm, but you could
keep your hand on it. (How's that for quantitative?)
I have two Z3801s and two Racal-Dana model 1992
counters set to Phase
A-B.
One counter has been measuring a 4040 cesium box
against one of the GPS
receivers. The other counter measures the phase
difference between the
Z3801s. With the 96 unit in place of the 4040, phase
rotation was 220
degrees in 20 minutes. The GPS comparison doesn't
rotate, but varies
between 50 and 70 degrees. I consider that an
indication that the
receivers
are OK. The holdover accuracy varies between 0.5 and
5 microseconds,
depending on when the crystal last cracked.
If I did the math right, a 360 rotation will take
2000 seconds. This
means
the LPRO is about 5x10E-11 off from GPS, no?
Two other questions:
What is the BITE pin on the LPRO? Some Built-In
Test?
What is the protocol for the data at J6? Can it be
the same as the
Z3801?
Bill Hawkins
time-nuts mailing list
time-nuts@febo.com
Fedora Core 4 Linux
Do You Yahoo!?
Tired of spam? Yahoo! Mail has the best spam protection around
http://mail.yahoo.com
Bill:
If you don't have it already, you might want to download the LPRO User
Manual:
http://www.symmetricom.com/media/pdf/manuals/man-lpro.pdf
BITE: Built-In Test Equipment (unlock indicator)
Jason
Two other questions:
What is the BITE pin on the LPRO? Some Built-In Test?
What is the protocol for the data at J6? Can it be the same as the
Z3801?
I know this is a little OT for this group, but I was looking for a
replacement knob for my Tek 2215 scope. Just one of the little gray ones
that populates a majority of the dials. I'm pretty sure the 22xx models are
all the same, possibly the entire 2000 series.
Would anyone here happen to have a pile of surplus knobs that would be
willing to sell me the right one if they have it? I could email a picture of
the exact knob I'm looking for if you need me to.
crosses fingers
Thanks guys,
Jason
Heck I would be grateful for a service manual for my Tektronix 2246. One doesn't
seem to exist.
-----Original Message-----
From: time-nuts-bounces@febo.com [mailto:time-nuts-bounces@febo.com] On Behalf
Of Jason Rabel
Sent: Wednesday, January 10, 2007 10:50 AM
To: 'Discussion of precise time and frequency measurement'
Subject: [time-nuts] Looking For Tektronix Knobs 22xx Series
I know this is a little OT for this group, but I was looking for a
replacement knob for my Tek 2215 scope. Just one of the little gray ones
that populates a majority of the dials. I'm pretty sure the 22xx models are
all the same, possibly the entire 2000 series.
Would anyone here happen to have a pile of surplus knobs that would be
willing to sell me the right one if they have it? I could email a picture of
the exact knob I'm looking for if you need me to.
crosses fingers
Thanks guys,
Jason
time-nuts mailing list
time-nuts@febo.com
https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts